Add task for volume analysis

arkindex/dataimport/management/commands/analyze_ml.py

+#!/usr/bin/env python3
+from django.core.management.base import BaseCommand, CommandError
+from arkindex.documents.models import Element, ElementType, Page
+from arkindex.dataimport.tasks import save_ml_results
+from arkindex.project.celery import app as celery_app
+
+
+class Command(BaseCommand):
+    help = 'Analyze a volume using worker-ml'
+
+    def add_arguments(self, parser):
+        parser.add_argument(
+            'volume',
+            help='Volume ID to analyze',
+        )
+
+    def handle(self, *args, **options):
+        # Lookup volume
+        try:
+            volume = Element.objects.get(id=options['volume'], type=ElementType.Volume)
+        except Element.DoesNotExist:
+            raise CommandError('Volume not found {}'.format(options['volume']))
+
+        # Build payload using pages
+        pages = Page.objects.filter(
+            pk__in=Element.objects.get_descending(volume.id, type=ElementType.Page)
+        )
+        payload = {
+            'volume': str(volume.id),
+            'pages': [
+                (page.id, page.zone.image.get_thumbnail_url(max_width=1500))
+                for page in pages
+            ]
+        }
+
+        # Start tasks with payload
+        analyze = celery_app.signature('arkindex_ml.tasks.analyze_pages')
+        workflow = analyze | save_ml_results.s()
+        task = workflow.delay(payload)
+        print('Started task {}'.format(task))